MS products are designed to function with as little user / administrator intervention as posible.
Ummmm, what MS product are you using? I have been in the industry over a decade now. I have worked as a sys admin, system architect, IT manager, data warehouse architect and technology consultant. And I can tell you that your assertion is simply not true! It typically takes about 2 times as many sys admins to administer a given number of MS servers as it would take for the equivalent number of proprietary UNIX or open source servers. This is a regurgitation of unsupported MS claims used to scare people (i.e. your total cost of ownership will be higher with Linux, offsetting the fact that it costs less than Win2K Server). It ain't true.
Now, you talk about the cost to the taxpayer, but when you're spending billions on things that blow up (where's the ROI in *that*?), that argument is shaky at best. I think the infrastructure could be refitted at the expense of a few less missles, while eliminating the secondary (Microsoft/Oracle/IBM) tax of proprietary software.
Since we are talking about the proposed law in California I think we can discard the idea that building a few less missiles will fund the refitting you're talking about. Even IF we were talking about the Federal government "a few less missiles" would not fund what you are talking about.
In any case, there are significant issues on the table with mandating open source software for the State of California. Before I go any further I should lay my cards on the table. I work for an IT services company. However, I'm expressing my own views, not the position of my employer. I currently work in one of the government divisions of that company, and for the last four years have supported multiple contracts with the state of California. I have a serious stake in this law, both because of my job and because I'm a citizen of California. Some of what I have to say deals with my political views as a citizen and some with my views as an IT professional.
I'm a huge supporter of open source. I think it's clear that the Internet, as we know it, was created by open source platforms, including Berkley UNIX (and subsequently BSD and Linux), BIND, and Apache. The open source community has been, and still is, a leader in many of the innovations in computing today. This same community responds much faster to customer needs, bugs and security holes than commercial vendors do. I wish I could get my commercial vendors to be as responsive.
That said, there are two major flaws with the idea of mandating open source only software in government IT. The first is that there are some things that open source simply cannot do. Perhaps in the future that will no longer be true, but it is not the case today. Some systems run by the state of California today could not run on Linux and MySQL due to their sheer size and complexity. In fact, at least one is still running on IBM mainframes because the risk involved in migrating to midrange platforms like IBM pSeries or Sun's SunFire is simply too high. While Linux can run on the mainframe, it cannot support the scope of this particular system, it is still Linux, running other open source platforms. There is some promising work being done in grid computing and super computing based on Linux that leads me to believe that this problem will be overcome in the next few years.
Politically, the correct approach to the issue of "transparency" for our government is three-fold:
Open standards rather than proprietary standards. For example, TCP/IP for network connectivity instead of SNA.
Open records for procurement, contracting and IT standards.
No proprietary data formats
In combination with the already existing freedom of information laws, this would ensure that, whether the IT platform is proprietary or open source it can interoperate with any other system using open standards. It would ensure that citizens have a full and informative view of the government's procurement process, allowing oversight to hopefully prevent something like the Oracle Master Licensing Agreement that California entered into last year. And finally it would ensure that government information is available to anyone with a web browser.
Mandating open source only is doomed to failure, at least at the stage of development of open source platforms that currently exist. There are many instances in state government where open source software could benefit the government and the taxpayer. On the desktop of office workers, as web servers, as office automation file and print servers, even replacing many of the proprietary systems in place today. And a mandate to include open source in procurement processes would help to make those changes. But a mandate to use only open source software will break California's IT systems.
Run a server and what else? Anything that linux (or win2000/xp) can't do?
Show me a win2k/xp server that supports thousands of users, accessing terabytes of information, 24X7 and doesn't go down...... ever.
Show me a linux distro that runs more than 8 CPU's, is supported by all the ISV's, can be run a cluster with 512 CPU's, and yet still runs well on a single CPU workstation.
Linux is a great workstation and small server choice. Win2K/XP is fine for the casual home user. But when you want to do real Enterprise computing then Solaris, AIX, OS/390, OS/400 or HP-UX is where you will end up at.
Actually, on Solaris you can create a package server and do a pkgadd -d , no problem. We use NFS file systems and do pkgadd's, works great, no fuss, no muss, one copy of the package on our management server. No harder to use than any other package program I've seen. Most GNU software that you could want has been built as a package by Sun Freeware already. Sun releases all their patches and add on software as packages. ISV's either release software as packages or create their own custom install interface (ala Oracle). I still don't see the issue, or why you think it's so hard. If you have an OEM install of Solaris, add libgcc, gcc and make to the server, it's highly unlikely that you will have any additional issues. Your path has to be correct of course, but that ought to be a no brainer.
I'm not quite sure what you are doing wrong. When I need gnu tools, I download the packages from Sun Freeware. If they don't have it (extremely unlikely) I download the source and compile it. I rarely, if ever, have problems doing that.
If you start by installing libgcc, gcc and make then compiling and installing source will be relatively simple. Hell, we use gcc for our third party apps that need a compiler, rather than buying Sun's C compiler.
For desktop use I agree that Linux is much superior to Solaris, although Sun is goint to move from CDE to Gnome for their GUI and that should improve things dramatically. For application support (especially databases) I cannot imagine that Linux would run better on SPARC architecture than Solaris does.
IF you own Sun Hardware AND have a service contract on your Solaris software, it costs nothing. Since people buying servers that big are paying a million plus for the hardware, they also purchase annual maintenance agreements from Sun. The cost of your Solaris maintenance agreement is $0 if combined with hardware maintenance.
In ten years of supporting Sun platforms I have yet to pay for a operating environment upgrade.
Sun Microsystems announced the Sun LX50 today at LinuxWorld. They also are again shipping Solaris 8 for x86, the cost is $45 for the CD or $20 to download the CD image. They have not yet released Solaris 9 for x86 for general availability, as far as I can tell.
The LX50 is intended for edge computing. For example as a web or proxy server. It runs "Sun Linux 5.0", although I can't find out much about what that distro is. It appears to be based on the UltraLinux distro.
Since Solaris 8 for x86 was one of the fastest and most stable UNIX releases for Intel platforms I would venture to guess that the 9 release will be also. The usual problem will be the lack of ISV applications for the product. Although numerous open source packages will be available, making it a great web server or email server.
Microsoft reliability with UNIX(tm) price. unbeatable.
Since I wouldn't pay $10 for so-called Microsoft reliability (the last MS OS I bought was Win95 because it was bundled with my PC) Xenix must be free???
Re:All OSS no better than all CSS
on
Mega-Geek March?
·
· Score: 1
You are correct. In order to run LPAR's on IBM pSeries or zSeries you have to have OS/390 (zSeries) or AIX (pSeries) running and then create an LPAR that is allowed to run Linux, the LPAR fakes Linux out to think that the hardware is compatible.
Sun is probably going to announce low end server platforms running Linux next week. This will be a native implementation. And Linux has been ported to SPARC architecture. So it's a matter of time now until Linux can run native on large Sun platforms. Since Sun using hardware partitioning technologies instead of LPAR technologies there won't be any issues with having to run Solaris 9 to enable Linux to run. If Sun really embraces Linux as the rumors say, this is a much more likely way to get Linux into Enterprise computing.
Re:All OSS no better than all CSS
on
Mega-Geek March?
·
· Score: 1
Political reasons are the key, at least in the services provided by IT contractors. When new systems are proposed the government contract manager wants to know that the system will work, be on budget, and on time, before you have even started the technical design. If what you are proposing does not match their current systems (e.g. the govt. dept. uses MS and SQL server and you are proposing Solaris and Oracle) then you will have a huge uphill battle to fight. "Independent" auditors are often brought in to review, validate and verify everything. Bottom line, if there are problems, either with budget, schedule or performance, the contract manager is in big political trouble within the government bureaucracy.
It's tough to sell alternative platforms to whatever is that department's mainstream, whether you want to replace mainframes with midrange, or MS with UNIX or UNIX with MS, or whatever. Even worse is selling them on Open Source. In the contract manager's view there won't be anyone to point to and place the blame when the system doesn't work. They can't protect themselves and that scares them. The auditor will tell them it's unproven technology, other contract managers will point out that when they used X OS and Y application it all worked, why use that Open Source stuff, and so forth.
I do know that the U.S. govt uses Linux and BSD for a lot of their platforms these days. I believe that it took a serious top down push for that to happen, although I'm not as familiar with the workings of US govt IT as I am with state of California.
California could save a lot of $$$ by transitioning from MS desktops, file servers, web servers, etc. to Open Source and still keep their big iron platforms in place where they are needed because Open Source isn't ready for those arenas. Will they do it? My guess is one of two things.
The bill mandating open source will be pass, and the implementation will be horrible and cost the taxpayers huge amounts of money
The bill won't get passed because the incumbent govt has already been burned way too many times recently by IT (even though it was their own fault for sole sourcing, taking bribes, etc.)
Yes, I'm aware that's pessimistic but that's what I have seen in nearly a decade of involvment with state IT.
Re:All OSS no better than all CSS
on
Mega-Geek March?
·
· Score: 3, Informative
Is there anything in particular that makes you shy of Linux on these big systems?
I'm currently a government IT contractor. I am an architect for Enterprise systems. I'm also a huge proponent of Open Source. OS/390, now called zOS, is incredibly stable and capable. Yes, it's proprietary, but it does the job. In one particular case that I'm deeply involved with we have a system that runs on OS/390 that provides benefits to 5 million Californians and deals with about $18 billion in financials annually. While the Linux kernel itself can handle that level of computing until recently I have not seen any applications, proprietary or open source, that ran on Linux that could support the system.
Secondarily to that I don't see the ability to actually run on the hardware platforms you need to support a system like that. We are talking about either IBM zSeries or pSeries or Sun SunFire level hardware. To the best of my knowledge Linux does not currently support SMP computing when 15, 20, 30, or more CPU's are involved on a single machine. Undoubtedly it will be able to, but it doesn't today.
By mandating Open Source you will eliminate some of the most stable and reliable RDBMS available (i.e. Oracle, DB2, Sybase) and force those same systems to use MySQL. Before someone gets their panties in a bunch, MySQL is a good database, but it's not yet capable of running multi-terabyte data warehouses, or transactional systems that support millions of transactions a day.
I think that, as of today, Open Source can replace desktop computers, especially for two arenas:
Office automation
Developers
Open Source can also replace all those proprietary web/ftp/mail/file/print servers and do a better job than most of them for a much better price. The Linux/SAMBA combo is pretty damn good. But, because the work hasn't yet been done, I don't think that Linux is ready for Enterprise level computing. Linux needs to be ported to those platforms and some work will have to be done on the kernel for it to work appropriately. The applications aren't there yet either and that will take longer.
Re:Geeks marching? Not bloodly likely.
on
Mega-Geek March?
·
· Score: 1
If, for instance, the government in its infinate wisdom decides to use Microsoft Word for its daily document needs then it should require that the documents are readily translatable in to an open source format.
I have worked on several government contracts. I currently am employed by a company that provides services to the state of california. When we provide information to the public or to customers of the state we are required to either provide it in PDF format OR to provide the means to view the information if it is not in PDF format. Providing a downloadable viewer for Word, for example. This is the norm throughout state IT contracts.
Public Domain licensing of software and data is a better bet than mandating Open Source.
Re:All OSS no better than all CSS
on
Mega-Geek March?
·
· Score: 1
You're right, but BSD, for example is on par with Linux in terms of Enterprise level capabilities. Yes IBM and Sun are putting emphasis on Open Source these days. But it's not there yet. Are we going to force the state to migrate all their big iron hardware running critical services to Open Source in the hopes that these vendors will somehow make Linux able to support that level of computing? And what will be the reaction when the DMV or Medi-Cal or whatever breaks?
Re:All OSS no better than all CSS
on
Mega-Geek March?
·
· Score: 2, Informative
**Laugh** No, like actually running a sysplex of several Z series servers with CICS on top of it, providing millions of transactions per day in the state's welfare system. Linux *can* run on IBM 390 hardware, yes. It can't do what OS/390 can do, or what Solaris on SunFire platforms can do.
Now, Linux will be able to provide Enterpise level computing in the future, but it's not there today. What I prefer is public domain licensing. In many state contracts the OS and apps may be proprietary but all the custom integration work is public domain and freely available to anyone.
Re:All OSS no better than all CSS
on
Mega-Geek March?
·
· Score: 1
Yes, I've read it.
But tell me, when this law is passed, is the state of California going to have to replace it's mainframe systems with little intel servers running Linux? Are the state's welfare systems (Medi-Cal, Healthy Families, etc.) Going to try and service millions of people and their health providers with Linux? Those systems are immmense. They might be able to rewrite and run on AIX or Solaris, but I don't think Linux can support mission critical big iron yet.
I had an interesting experience just this morning. I have been a firm believer in a few things:
Proprietary UNIX operating environments like Solaris and AIX are, in all aspects, the best platform for back office services... i.e. data warehousing, CRM, etc.
Linux based operating environments are the best platform for front tier platforms like web servers, app servers, file & print, etc. (especially using SAMBA, RH puts Win2K to shame!)
Windows, especially Win2K (XP isn't worth the $$$ to upgrade) is the best choice for the typical office user and home user.
But, I've been wanting to try out Linux on a PC and see how much it's improved at the consumer level. So, I installed RedHat 7.3 on my IBM T20 laptop. I've had this laptop for nearly 2 years. I have a CD I built with all necessary Windows 2000 drivers on it. The machine came with Win98 and I didn't feel like getting IBM's Win2K, so I just used a generic Win2K installation that I own. I had to download about 20 IBM specific drivers and install them before the laptop worked "right". It functioned, but not well.
So, I downloaded RH 7.3 iso's from NASA (blazing download speeds, over 1.7 Mbps) and burned the install CD's. I then popped CD #1 in my laptop and rebooted. In less than 1 hour my laptop was a functional dual-boot machine. I let RedHat make all the install decisions, rather than customizing like I would do on a server. I allowed GRUB to be my boot loader. It boots both Win2K and RH beautifully, no issues. RH runs great AND I didn't have to download one single driver to get my system to work with Linux. Win2K doesn't include support for my 2 year old network card, so I have to have that driver downloaded before attempting a Win2K install on this laptop, no such problem with RH.
I'm a believer now. RH 7.3 is definitely ready for the average end user's PC. The installation is no more difficult than Windows, if you set it to boot to graphical logon mode life is easy. And once in Gnome (or KDE) all the tools that a typical end user might want are there. With about the same amount of effort that it takes to install Office XP that same user can download, install and use Open Office (that took me about 30 minutes).
Best of all, I didn't have to use knowledge gained in 10 years of implementing and administering UNIX servers. It was pretty damn easy. To get the same easy installation with Win2K on a laptop I need to get the OEM version of Win2K appropriate to my laptop OR I can just get the generic RH distro. No issues, no worries, no compatibility problems.
Bottom line, MS OS is no longer superior in the consumer market based on what I just saw, and the Linux price is hard to beat. For the user who doesn't want to deal with creating their own CD's they can pick up the boxed set of RH for a low price down at Best Buy. Within an hour they can have a functioning system that is equal to Windows. The only thing keeping it from going mainstream is games. Come on game developers. Get those games running on Linux and MS is in big trouble.
But who controls that centralized authentication database? How do I know that my private information is secure? What if that centralized source uses strong arm tactics to force web sites to use it? Will my privacy be protected? Whether it's Microsoft, or any other organization, these issues are present. That's the real problem with Passport.
There are no license costs for Solaris. Solaris RTU license is free for 8 CPU's or less. All Sun platforms include a RTU license regardless of number of CPU's. The "cost of Solaris" is FUD. Now, if you are saying that they have heartburn with the cost of big iron 64-bit computing platforms, that's different. Since both IBM and Sun handle their RISC servers the same way (i.e. you buy the hardware and the Operating Environment RTU is free) this won't change if you want to use Linux on big iron. The real cost savings is not at the high end, at least not yet. It's at the low end where I can get a 4 CPU x86 server, put Linux on it, and compete with a 2 to 4 CPU Sun server, for about half the price. I work for a Fortune 500 company and we don't "blanch at the license costs". The cost of high end computing platforms is part of the cost of doing business and is built into our budgets and cost models.
Heh, the last time I called MS for support (my company is an Enterprise partner of MS, and a Fortune 500 member) when I told them what company I worked for the MS guy responded "who?". He then proceeded to be clueless about his own product, I knew more about it than he did. I have never again called MS for support. I do use mailing lists, usenet, user groups and phone support for my UNIX, Linux, ISV and Open Source software, very successfully.
That is absolutely true and I should have mentioned that. However, I don't really need support for each PC I install an OS on, nor do I need support for my general purpose, non-mission critical servers. I can just install RedHat or Mandrake or whatever on them and support them myself. And with the availability of no cost support via mailing lists, usenet, user groups, not to mention the Open Source development teams, I generally get faster, better support than I do from Microsoft.
Mission critical servers are a different story. Although 10, or even 5, years ago, I was running web, dns and ftp services using Apache and BIND on Linux or SCO without support, and those were certainly mission critical to the organization I worked for. Support from a commercial entity is FUD, whether Microsoft or Sun or IBM is the one saying it.
In fact, Sun is primarily a hardware company. They do sell software, but they are not a market leader in any software they sell. Their bread and butter is servers and storage. Sun knows it and so does the Enterprise IT industry. If you buy a Sun server you get the OS RTU for free.
Slashdot Mirror
I said that to my wife the other night. It's a scary name for a government organization. She kind of shrugged and went on about her business.
MS products are designed to function with as little user / administrator intervention as posible.
Ummmm, what MS product are you using? I have been in the industry over a decade now. I have worked as a sys admin, system architect, IT manager, data warehouse architect and technology consultant. And I can tell you that your assertion is simply not true! It typically takes about 2 times as many sys admins to administer a given number of MS servers as it would take for the equivalent number of proprietary UNIX or open source servers. This is a regurgitation of unsupported MS claims used to scare people (i.e. your total cost of ownership will be higher with Linux, offsetting the fact that it costs less than Win2K Server). It ain't true.
Now, you talk about the cost to the taxpayer, but when you're spending billions on things that blow up (where's the ROI in *that*?), that argument is shaky at best. I think the infrastructure could be refitted at the expense of a few less missles, while eliminating the secondary (Microsoft/Oracle/IBM) tax of proprietary software.
Since we are talking about the proposed law in California I think we can discard the idea that building a few less missiles will fund the refitting you're talking about. Even IF we were talking about the Federal government "a few less missiles" would not fund what you are talking about.
In any case, there are significant issues on the table with mandating open source software for the State of California. Before I go any further I should lay my cards on the table. I work for an IT services company. However, I'm expressing my own views, not the position of my employer. I currently work in one of the government divisions of that company, and for the last four years have supported multiple contracts with the state of California. I have a serious stake in this law, both because of my job and because I'm a citizen of California. Some of what I have to say deals with my political views as a citizen and some with my views as an IT professional.
I'm a huge supporter of open source. I think it's clear that the Internet, as we know it, was created by open source platforms, including Berkley UNIX (and subsequently BSD and Linux), BIND, and Apache. The open source community has been, and still is, a leader in many of the innovations in computing today. This same community responds much faster to customer needs, bugs and security holes than commercial vendors do. I wish I could get my commercial vendors to be as responsive.
That said, there are two major flaws with the idea of mandating open source only software in government IT. The first is that there are some things that open source simply cannot do. Perhaps in the future that will no longer be true, but it is not the case today. Some systems run by the state of California today could not run on Linux and MySQL due to their sheer size and complexity. In fact, at least one is still running on IBM mainframes because the risk involved in migrating to midrange platforms like IBM pSeries or Sun's SunFire is simply too high. While Linux can run on the mainframe, it cannot support the scope of this particular system, it is still Linux, running other open source platforms. There is some promising work being done in grid computing and super computing based on Linux that leads me to believe that this problem will be overcome in the next few years.
Politically, the correct approach to the issue of "transparency" for our government is three-fold:
- Open standards rather than proprietary standards. For example, TCP/IP for network connectivity instead of SNA.
- Open records for procurement, contracting and IT standards.
- No proprietary data formats
In combination with the already existing freedom of information laws, this would ensure that, whether the IT platform is proprietary or open source it can interoperate with any other system using open standards. It would ensure that citizens have a full and informative view of the government's procurement process, allowing oversight to hopefully prevent something like the Oracle Master Licensing Agreement that California entered into last year. And finally it would ensure that government information is available to anyone with a web browser.Mandating open source only is doomed to failure, at least at the stage of development of open source platforms that currently exist. There are many instances in state government where open source software could benefit the government and the taxpayer. On the desktop of office workers, as web servers, as office automation file and print servers, even replacing many of the proprietary systems in place today. And a mandate to include open source in procurement processes would help to make those changes. But a mandate to use only open source software will break California's IT systems.
Run a server and what else? Anything that linux (or win2000/xp) can't do?
Show me a win2k/xp server that supports thousands of users, accessing terabytes of information, 24X7 and doesn't go down ...... ever.
Show me a linux distro that runs more than 8 CPU's, is supported by all the ISV's, can be run a cluster with 512 CPU's, and yet still runs well on a single CPU workstation.
Linux is a great workstation and small server choice. Win2K/XP is fine for the casual home user. But when you want to do real Enterprise computing then Solaris, AIX, OS/390, OS/400 or HP-UX is where you will end up at.
Actually, on Solaris you can create a package server and do a pkgadd -d , no problem. We use NFS file systems and do pkgadd's, works great, no fuss, no muss, one copy of the package on our management server. No harder to use than any other package program I've seen. Most GNU software that you could want has been built as a package by Sun Freeware already. Sun releases all their patches and add on software as packages. ISV's either release software as packages or create their own custom install interface (ala Oracle). I still don't see the issue, or why you think it's so hard. If you have an OEM install of Solaris, add libgcc, gcc and make to the server, it's highly unlikely that you will have any additional issues. Your path has to be correct of course, but that ought to be a no brainer.
If you start by installing libgcc, gcc and make then compiling and installing source will be relatively simple. Hell, we use gcc for our third party apps that need a compiler, rather than buying Sun's C compiler.
For desktop use I agree that Linux is much superior to Solaris, although Sun is goint to move from CDE to Gnome for their GUI and that should improve things dramatically. For application support (especially databases) I cannot imagine that Linux would run better on SPARC architecture than Solaris does.
What version of Solaris are you replacing?
Do you have a URL for that?
IF you own Sun Hardware AND have a service contract on your Solaris software, it costs nothing. Since people buying servers that big are paying a million plus for the hardware, they also purchase annual maintenance agreements from Sun. The cost of your Solaris maintenance agreement is $0 if combined with hardware maintenance.
In ten years of supporting Sun platforms I have yet to pay for a operating environment upgrade.
Why would you remove Solaris from a Sun box??? Is there an advantage I'm missing?
Sun Microsystems announced the Sun LX50 today at LinuxWorld. They also are again shipping Solaris 8 for x86, the cost is $45 for the CD or $20 to download the CD image. They have not yet released Solaris 9 for x86 for general availability, as far as I can tell.
The LX50 is intended for edge computing. For example as a web or proxy server. It runs "Sun Linux 5.0", although I can't find out much about what that distro is. It appears to be based on the UltraLinux distro.
Since Solaris 8 for x86 was one of the fastest and most stable UNIX releases for Intel platforms I would venture to guess that the 9 release will be also. The usual problem will be the lack of ISV applications for the product. Although numerous open source packages will be available, making it a great web server or email server.
Microsoft reliability with UNIX(tm) price. unbeatable.
Since I wouldn't pay $10 for so-called Microsoft reliability (the last MS OS I bought was Win95 because it was bundled with my PC) Xenix must be free???
You are correct. In order to run LPAR's on IBM pSeries or zSeries you have to have OS/390 (zSeries) or AIX (pSeries) running and then create an LPAR that is allowed to run Linux, the LPAR fakes Linux out to think that the hardware is compatible.
Sun is probably going to announce low end server platforms running Linux next week. This will be a native implementation. And Linux has been ported to SPARC architecture. So it's a matter of time now until Linux can run native on large Sun platforms. Since Sun using hardware partitioning technologies instead of LPAR technologies there won't be any issues with having to run Solaris 9 to enable Linux to run. If Sun really embraces Linux as the rumors say, this is a much more likely way to get Linux into Enterprise computing.
Political reasons are the key, at least in the services provided by IT contractors. When new systems are proposed the government contract manager wants to know that the system will work, be on budget, and on time, before you have even started the technical design. If what you are proposing does not match their current systems (e.g. the govt. dept. uses MS and SQL server and you are proposing Solaris and Oracle) then you will have a huge uphill battle to fight. "Independent" auditors are often brought in to review, validate and verify everything. Bottom line, if there are problems, either with budget, schedule or performance, the contract manager is in big political trouble within the government bureaucracy.
It's tough to sell alternative platforms to whatever is that department's mainstream, whether you want to replace mainframes with midrange, or MS with UNIX or UNIX with MS, or whatever. Even worse is selling them on Open Source. In the contract manager's view there won't be anyone to point to and place the blame when the system doesn't work. They can't protect themselves and that scares them. The auditor will tell them it's unproven technology, other contract managers will point out that when they used X OS and Y application it all worked, why use that Open Source stuff, and so forth.
I do know that the U.S. govt uses Linux and BSD for a lot of their platforms these days. I believe that it took a serious top down push for that to happen, although I'm not as familiar with the workings of US govt IT as I am with state of California.
California could save a lot of $$$ by transitioning from MS desktops, file servers, web servers, etc. to Open Source and still keep their big iron platforms in place where they are needed because Open Source isn't ready for those arenas. Will they do it? My guess is one of two things.
- The bill mandating open source will be pass, and the implementation will be horrible and cost the taxpayers huge amounts of money
- The bill won't get passed because the incumbent govt has already been burned way too many times recently by IT (even though it was their own fault for sole sourcing, taking bribes, etc.)
Yes, I'm aware that's pessimistic but that's what I have seen in nearly a decade of involvment with state IT.Is there anything in particular that makes you shy of Linux on these big systems?
I'm currently a government IT contractor. I am an architect for Enterprise systems. I'm also a huge proponent of Open Source. OS/390, now called zOS, is incredibly stable and capable. Yes, it's proprietary, but it does the job. In one particular case that I'm deeply involved with we have a system that runs on OS/390 that provides benefits to 5 million Californians and deals with about $18 billion in financials annually. While the Linux kernel itself can handle that level of computing until recently I have not seen any applications, proprietary or open source, that ran on Linux that could support the system.
Secondarily to that I don't see the ability to actually run on the hardware platforms you need to support a system like that. We are talking about either IBM zSeries or pSeries or Sun SunFire level hardware. To the best of my knowledge Linux does not currently support SMP computing when 15, 20, 30, or more CPU's are involved on a single machine. Undoubtedly it will be able to, but it doesn't today.
By mandating Open Source you will eliminate some of the most stable and reliable RDBMS available (i.e. Oracle, DB2, Sybase) and force those same systems to use MySQL. Before someone gets their panties in a bunch, MySQL is a good database, but it's not yet capable of running multi-terabyte data warehouses, or transactional systems that support millions of transactions a day.
I think that, as of today, Open Source can replace desktop computers, especially for two arenas:
- Office automation
- Developers
Open Source can also replace all those proprietary web/ftp/mail/file/print servers and do a better job than most of them for a much better price. The Linux/SAMBA combo is pretty damn good. But, because the work hasn't yet been done, I don't think that Linux is ready for Enterprise level computing. Linux needs to be ported to those platforms and some work will have to be done on the kernel for it to work appropriately. The applications aren't there yet either and that will take longer.If, for instance, the government in its infinate wisdom decides to use Microsoft Word for its daily document needs then it should require that the documents are readily translatable in to an open source format.
I have worked on several government contracts. I currently am employed by a company that provides services to the state of california. When we provide information to the public or to customers of the state we are required to either provide it in PDF format OR to provide the means to view the information if it is not in PDF format. Providing a downloadable viewer for Word, for example. This is the norm throughout state IT contracts.
Public Domain licensing of software and data is a better bet than mandating Open Source.
You're right, but BSD, for example is on par with Linux in terms of Enterprise level capabilities. Yes IBM and Sun are putting emphasis on Open Source these days. But it's not there yet. Are we going to force the state to migrate all their big iron hardware running critical services to Open Source in the hopes that these vendors will somehow make Linux able to support that level of computing? And what will be the reaction when the DMV or Medi-Cal or whatever breaks?
**Laugh** No, like actually running a sysplex of several Z series servers with CICS on top of it, providing millions of transactions per day in the state's welfare system. Linux *can* run on IBM 390 hardware, yes. It can't do what OS/390 can do, or what Solaris on SunFire platforms can do.
Now, Linux will be able to provide Enterpise level computing in the future, but it's not there today. What I prefer is public domain licensing. In many state contracts the OS and apps may be proprietary but all the custom integration work is public domain and freely available to anyone.
Yes, I've read it.
But tell me, when this law is passed, is the state of California going to have to replace it's mainframe systems with little intel servers running Linux? Are the state's welfare systems (Medi-Cal, Healthy Families, etc.) Going to try and service millions of people and their health providers with Linux? Those systems are immmense. They might be able to rewrite and run on AIX or Solaris, but I don't think Linux can support mission critical big iron yet.
I had an interesting experience just this morning. I have been a firm believer in a few things:
- Proprietary UNIX operating environments like Solaris and AIX are, in all aspects, the best platform for back office services
... i.e. data warehousing, CRM, etc.
- Linux based operating environments are the best platform for front tier platforms like web servers, app servers, file & print, etc. (especially using SAMBA, RH puts Win2K to shame!)
- Windows, especially Win2K (XP isn't worth the $$$ to upgrade) is the best choice for the typical office user and home user.
But, I've been wanting to try out Linux on a PC and see how much it's improved at the consumer level. So, I installed RedHat 7.3 on my IBM T20 laptop. I've had this laptop for nearly 2 years. I have a CD I built with all necessary Windows 2000 drivers on it. The machine came with Win98 and I didn't feel like getting IBM's Win2K, so I just used a generic Win2K installation that I own. I had to download about 20 IBM specific drivers and install them before the laptop worked "right". It functioned, but not well.So, I downloaded RH 7.3 iso's from NASA (blazing download speeds, over 1.7 Mbps) and burned the install CD's. I then popped CD #1 in my laptop and rebooted. In less than 1 hour my laptop was a functional dual-boot machine. I let RedHat make all the install decisions, rather than customizing like I would do on a server. I allowed GRUB to be my boot loader. It boots both Win2K and RH beautifully, no issues. RH runs great AND I didn't have to download one single driver to get my system to work with Linux. Win2K doesn't include support for my 2 year old network card, so I have to have that driver downloaded before attempting a Win2K install on this laptop, no such problem with RH.
I'm a believer now. RH 7.3 is definitely ready for the average end user's PC. The installation is no more difficult than Windows, if you set it to boot to graphical logon mode life is easy. And once in Gnome (or KDE) all the tools that a typical end user might want are there. With about the same amount of effort that it takes to install Office XP that same user can download, install and use Open Office (that took me about 30 minutes).
Best of all, I didn't have to use knowledge gained in 10 years of implementing and administering UNIX servers. It was pretty damn easy. To get the same easy installation with Win2K on a laptop I need to get the OEM version of Win2K appropriate to my laptop OR I can just get the generic RH distro. No issues, no worries, no compatibility problems.
Bottom line, MS OS is no longer superior in the consumer market based on what I just saw, and the Linux price is hard to beat. For the user who doesn't want to deal with creating their own CD's they can pick up the boxed set of RH for a low price down at Best Buy. Within an hour they can have a functioning system that is equal to Windows. The only thing keeping it from going mainstream is games. Come on game developers. Get those games running on Linux and MS is in big trouble.
You would only be normal in MY universe if you were me. Are you me? Am I you? Maybe I'm just a dream and when you wake up I'll go away?
But who controls that centralized authentication database? How do I know that my private information is secure? What if that centralized source uses strong arm tactics to force web sites to use it? Will my privacy be protected? Whether it's Microsoft, or any other organization, these issues are present. That's the real problem with Passport.
There are no license costs for Solaris. Solaris RTU license is free for 8 CPU's or less. All Sun platforms include a RTU license regardless of number of CPU's. The "cost of Solaris" is FUD. Now, if you are saying that they have heartburn with the cost of big iron 64-bit computing platforms, that's different. Since both IBM and Sun handle their RISC servers the same way (i.e. you buy the hardware and the Operating Environment RTU is free) this won't change if you want to use Linux on big iron. The real cost savings is not at the high end, at least not yet. It's at the low end where I can get a 4 CPU x86 server, put Linux on it, and compete with a 2 to 4 CPU Sun server, for about half the price. I work for a Fortune 500 company and we don't "blanch at the license costs". The cost of high end computing platforms is part of the cost of doing business and is built into our budgets and cost models.
Heh, the last time I called MS for support (my company is an Enterprise partner of MS, and a Fortune 500 member) when I told them what company I worked for the MS guy responded "who?". He then proceeded to be clueless about his own product, I knew more about it than he did. I have never again called MS for support. I do use mailing lists, usenet, user groups and phone support for my UNIX, Linux, ISV and Open Source software, very successfully.
That is absolutely true and I should have mentioned that. However, I don't really need support for each PC I install an OS on, nor do I need support for my general purpose, non-mission critical servers. I can just install RedHat or Mandrake or whatever on them and support them myself. And with the availability of no cost support via mailing lists, usenet, user groups, not to mention the Open Source development teams, I generally get faster, better support than I do from Microsoft.
Mission critical servers are a different story. Although 10, or even 5, years ago, I was running web, dns and ftp services using Apache and BIND on Linux or SCO without support, and those were certainly mission critical to the organization I worked for. Support from a commercial entity is FUD, whether Microsoft or Sun or IBM is the one saying it.
In fact, Sun is primarily a hardware company. They do sell software, but they are not a market leader in any software they sell. Their bread and butter is servers and storage. Sun knows it and so does the Enterprise IT industry. If you buy a Sun server you get the OS RTU for free.